APR: your source for nuclear news and analysis since April 16, 2010

Tuesday, April 24, 2012

San Onofre and Steam Generator Design

Readers of Atomic Power Review have certainly noticed some focus on the continuing situation at SoCal Ed's San Onofre nuclear station in the last few weeks, due to the problems experienced with the replacement steam generators installed in that plant. APR has made a number of posts on this topic, as reports and as official information have warranted.

Because of the rather intense media coverage and also because of the deliberate focusing of energy on this matter by established anti-nuclear forces, I wrote a considerably detailed piece on steam generator design for APR which can be seen by clicking here; this piece appeared first on March 20, 2012 and has had considerable traffic.

Now that further details about the real problems being encountered at San Onofre 2 and 3 have been released, and since the anti-nuclear forces at large have not let off their campaign of fear, a number of us have decided that it was time to present something substantial in order to let our regular readers here know what is really going on "inside" in terms of the steam generator design, and concerning the replacement.

For this, I've been very fortunate to be able to ask Meredith Angwin to construct a piece covering just those details. Meredith not only is well known as the author of Yes Vermont Yankee; she was also closely involved with steam generator design and secondary chemistry work during her years working with the Electric Power Research Institute. Meredith wrote the paper you are about to read and submitted it to a number of industry experts and analysts for peer review, which has taken some time. Today, the piece is prepared for publication - and with this publication I will also note that Meredith is the first ever guest author of a piece on Atomic Power Review. Her presentation and her followup comments are now presented; I will have some closing remarks at the end of this post.



San Onofre Steam Generators

In recent days, the tubes at the steam generators at the San Onofre Nuclear Generators 2 and 3 (SONGS 2 and SONGS 3) were discovered to have unusual wear patterns. These steam generators were replaced recently (circa 2010) and such problems were not expected. The plant owners and NRC have agreed to take the plants out of service while the root cause analysis of the wear is investigated. (APR note: Click here for a link to Mitsubishi's page on the replacement steam generators.)

Gundersen Reviews the Generators

Meanwhile, Arnie Gundersen was paid by the Friends of the Earth (FOE) to write a report about these generators and their problems. As a matter of fact, Gundersen has written two reports

Gundersen (Fairewinds) report 1

Gundersen (Fairewinds) report 2

Much of Gundersen’s report appears to be based on an article in Nuclear Engineering International magazine about the development of the SONGS generators. The report was written about the generator design process, before any issues were found at the generators themselves.

Nuclear Engineering International S/G article (relevant pages) Courtesy Will Dalrymple / NEI

The NEI magazine article’s insight into the design process allows Gundersen to put an interesting negative spin on things. In his first report, Gundersen’s main assertion is that the SONGS steam generators were replaced with new generators that were not exactly like the old generators, and this was the cause of the problem. He claimed that they changed the alloy, changed the design, and the new configurations were not tested. Also somehow all this sneaked by the NRC without review, because the new steam generators were supposed to be just like the replaced steam generators.

In Gundersen's second report, he moderated his stance a little bit. He admitted that alloy 690 is a better alloy for steam generator tubing than alloy 600. However, in the second report, he continued to say that except for the alloy, the changes in the new generators were “unreviewed design changes” that probably caused the problems. He was particularly concerned with the tubesheet “failing” and realizing radiation “directly into the environment.”

Electric Power Research Institute (EPRI) and me

Instead of dissecting Gundersen's reports word for word, I want to show that no steam generator will ever be replaced by a steam generator exactly like the old one. Extensive research has been done to improve steam generator design and performance. I will discuss research as it was led by the Steam Generator Owner's Group at EPRI in the mid-80s to the 90s.

I was a project manager and then a consultant to that group from 1982 to the late 1990s. My speciality was water chemistry, corrosion control, and testing of alloys against corrosion. I have not followed the continued Steam Generator changes since around 1998-2000, when my career took a different path. However, I have had this article reviewed for accuracy by several people. (Any mistakes are all my own.) This article will show how research has continuously improved steam generators. A new steam generator will never be just like the older version.

Thank heavens!

Starting with the Tubesheet

The tubesheet is a thick piece of metal that sits at the bottom of the steam generator and separates the primary side (water that flows through the reactor) from the secondary side (water that flows through the turbines). Holes are drilled into the tube sheet, and the steam generator tubes are inserted into those holes. Primary water flows inside of the tubes, and secondary water flows outside of the tubes. The tubes are the source of the heat transfer.

Above: Tube sheet forging for nuclear steam generator; illustration courtesy China Heavy Machinery Industry Association.

Above: Horizontal boring machine performing drilling operation on nuclear plant heat exchanger tube sheet at Babcock & Wilcox plant, Barberton, Ohio. Photo from APR collection / Will Davis.

(Note: if the tubesheet “failed”—an event that has never occurred—it would mix the primary and secondary waters. It would not “release radioactivity into the environment” as Gundersen described it. The secondary water is not “the environment.”)

According to the article in NEI, San Onofre’s planning for the tube sheet replacement included building a full-scale mock-up. I have don’t know about the testing they did, but I will talk a little about tube sheets and what we did at EPRI.

First, you have to drill a hole in the tubesheet. Once the holes are (carefully) drilled, the tubes must be inserted. The hole is bigger than the tube (of course) but the tube must make a good seal with the hole in order for the tube sheet to remain a pressure barrier. Something (an expander, a small explosive charge, various methods) is placed within the tube, and it is pushed out forcefully until it is really sealed into the tubesheet. This sealing process is only for the lower part of the tube within the thick tubesheet. Near the top, the tube is not sealed within the tube sheet, but an open crevice exists.

Tube Sheet Issues

Sealing the tube into the tubesheet involves cold-work of a metal, which introduces strains into the tube. In general, the next step is to relieve strains by some type of annealing/heating method of the tube. Both expansion methods and annealing methods were active research areas at EPRI (though not my area.)

The crevice is there because otherwise the tube would go from tightly-sealed in place within the tubesheet to moving around in the water flow. This sudden transition could break the tube. The crevice sort of supports the tube and eases the transition. However, the crevice is a corrosion trap. More on that later.

The Tubes and the Alloys

Immense amounts of work was done, for years, on proper alloys and heat treatments for the tubing. After extensive testing, Alloy 690 with a certain heat-treatment (nicknamed sensitisation, of all things) was found to be most resistant to stress corrosion cracking, which is usually the worst problem the tubing encounters. The tube usually encounters this problem (surprise) at the tubesheet crevice. The work on alloy 690 included much testing in scale models of all types, and also careful assessment of existing tubes (alloy 600) and their problems, figuring out what heats and heat treatments seemed best able to cope with corrosion.

The Support Plates

Tube support plates are a mixed blessing. They are large disks (with holes for the tubes) that are at various points in the steam generator. They are necessary to prevent the tubes from moving around too freely in the flow of secondary water. However, they are also something that can fret the tubes that bang into them, and the support plates themselves corrode. Support plate corrosion products are more voluminous than the support plates were, and they fill the space between the support plates and the tube. Then the corrosion products continue to grow, eventually pushing against the tubes, denting them and causing them to fail routine inspections and be taken out of service (plugged).

Above: Steam generator support plates. Photo courtesy AREVA via Meredith Angwin.

As I said, support plates are a mixed blessing. So it's always a tradeoff (engineering is like that) about how many support plates you put in...vibration versus tube denting. Take your compromise here!

The first generation of support plates can most simply be described as awful. The holes for the tubes were drilled or punched with no particular quality control on the size. The material was a carbon steel that corroded like crazy, denting tubes right and left. When I saw the new support plates being made by Areva in France with the careful measurements, polishing, 304 stainless steel...I was so impressed! Nobody would buy a steam generator like the "old days" anymore!

Crevices and Modeling

Okay, let's get back to those crevices. We will talk about the crevice in the tubesheet, and the somewhat similar crevices between the tubes and the support plates, higher in the steam generator. This means we also have to talk about water chemistry.
When I joined the Steam Generator Owner's Group, I came from the geothermal energy group, and figured that with the usual water control of the water that went through the turbine (secondary water), the steam generators would have no problems of the kind we had in geothermal. In geothermal, we had to use whatever water nature chose to supply and many times it was brine. We sometimes used a shorthand of 2X, 3X etc for twice as concentrated as seawater, three times as concentrated etc.

Hello steam generators. Same shorthand for the water in the crevices!

The temperature differential in the crevices concentrates salts, and the concentrated salts are very corrosive. The problem is simple. The crevices have poor heat transfer so they are basically at the temperature of the primary side. The rest of the steam generator is at the lower temperature of the secondary side. However, the whole steam generator has to be at the same pressure.
The crevices are hotter than the bulk water. Their higher temperature should mean that the crevices would have a higher pressure than the bulk water. But they can’t have a higher pressure than the bulk water.

So what happens? The presence of ions in the water lowers the pressure of the water, at any given temperature. The hotter crevices end up collecting ions, which lowers the partial pressure of the water in them to the same pressure as the water in the rest of the generator.

If you take a sample of bulk water in feedwater or blowdown, it might be 50 ppb ions. Meanwhile, back in the crevice--2X! The crevice has ions at twice the concentration of seawater. So the crevice has same pressure as the bulk water, though it is at a higher temperature.

This also led to the early belief that "steam generators clean themselves up." The idea is that the bulk water might be 20 ppm salts when you start up the SG , but by the time you are at temperature, bulk water would be at 1 ppm. A miracle! Conservation of mass finally defeated! Of course, the ions were "hiding out" in the crevices, and would "return" when the temperature dropped. People eventually caught on, but crevice corrosion of all sorts were very common in the early days of supposedly "self-cleaning" steam generators.

By the time I joined the Steam Generator group at EPRI, people knew about hideout and hideout return. The push to clean up the incoming water was in full swing. Water impurities were cleaned, and incoming water was more likely to be at 50 ppb than 50 ppm. I was involved with that, and I even recruited ultrapure water specialists from the Silicon Valley semiconductor industry as consultants to the utilities.

Alas, it was only partially helpful. Cleaning the water helps slow down the concentration process in the crevice, but it doesn't stop it. The physical requirement for equal pressure throughout the steam generator trumps any pathetic attempt to slow down the process of concentration. We had to try to understand the crevice chemistry, and add ions to the steam generator that would neutralize the crevice and stop most of the corrosion.

At a simple level, we knew that the source of the small levels of contamination in the water was miniscule inleakage from the condenser. If the plant was on freshwater, the inleakage from the condensers concentrated basic, leading to alkaline types of corrosion. Seawater inleakage concentrated acid. At a place like Indian Point, where the river is tidal (sometimes fresh, sometimes salt) and either type of water could cause the inleakage, all bets were off.

We tried modeling the crevices, and that was only partially successful. The problem is that most water models depend on the Debye-Huckel assumptions of a relatively dilute solution.


When you have a very concentrated solution (2X, 3X) and high temperatures (water is less polar at high temperatures), the basic science simply isn't there for modeling.

Crevices and Expensive Experiments

What should we to do for experiments, since modeling was of very limited help? Well, it got expensive. When we worked with denting, we could use "pots"...that is, lab-bench high pressure autoclaves, to do the corrosion testing for the tube support plate material. We could use lesser concentrations than true crevice chemistry and get good results.

When testing stress corrosion cracking and so forth on alloy 600 and 690 tubes in crevices--we had to go all the way to true crevice chemistry or nothing bad would happen. However, no ordinary equipment would handle those temperatures and high concentrations in bulk. Bench-top autoclaves were close to useless. Also, we didn't know what the true chemistry was. So we used model boilers, with real heat transfer and real crevices, and tested that way. This was madly expensive.

Model boilers at Westinghouse were single tube mock-ups of steam generators. The primary water side was heated electrically, not by fission, but everything else was the same as a steam generator. We made real progress on proper water chemistry for alloy 600 (existing steam generators) and alloy 690 (new generators). Another help was getting all the copper and admiralty brass out of the water systems, and changing out admiralty brass condensers for titanium at seawater plants. It takes very little copper to add a lot of OOMPH to the corrosion in crevices.

Steam Generators Improve

So much progress was made by this research at EPRI while I was there! Alloy choice. Heat treatment choice. Fabrication of tubes. Stress relief of the tubes in the tube sheets. Tube support plates: Hole size, hole drilling techniques, material choice, quality control. Water chemistry, tested under the exact conditions of a real steam generator (model boilers) at great expense.
Model boilers were huge (several stories tall) and were NOT cheap to run. Also, model boiler experiments often lasted for months. (How to go through a lot of money very fast.) Sometimes, we all thought this was annoying, because there were many possible bench-scale experiments that could have been done for the same money. But...bench scale didn't give us heat transfer and crevice chemistry. End of story.


I am a chemist, and I didn't work on the fretting and wear and vibration side of steam generators. For these issues, computer modeling seemed to be very helpful (unlike crevice chemistry modeling), and there were experiments, including some in the model boilers.

Now, you may say that I am being prejudiced in favor of the difficult issues chemists must face, and scornful of other people's problems. Okay. Say it. But still..

Vibration issues depend on many factors, including geometry and flow. Which also means that vibration patterns are relatively easy to change. There's nothing in vibration issues as tough as the ions-accumulate-in-the-crevice phenomenon. Most vibration issues are subject to amelioration without changing the laws of nature. Many things change the vibrational pattern of a steam generator. In contrast, you would have to change the laws of nature to keep crevices from accumulating salt loadings.

Angels Won't Tread

And now, boldly stepping where angels fear to tread on Root Cause analysis for San Onofre...no matter how long a list of changes Gundersen claims for the new steam generators there, in my opinion, corrosion will not turn out to be any part of the problem. The root cause analysis will come up with something vibration connected.

To cure it, some kind of derating of water flow (primary side and/or secondary side) with ameliorate the problem well enough to keep the generators in operation while the owners and the manufacturers sue each other. The generators will be replaced early. In the Steam Generator Owner's Group, we had a mantra of setting everything up to be "leak before break" (no catastrophic failures). That will still be the case for SONGS.

Nobody would buy an old-style steam generator nowadays, and that is a good thing! There's no bait-and-switch about putting in a new style steam generator.

The SONGS will rise again.


Meredith adds the following: "I am grateful for the knowledgable people who reviewed this article. They reviewed it on their own time, but they work for big companies. Therefore, for everyone's happiness, they will not be named here. I thank them. Luckily, I can thank Dan Yurman publicly for his review! Any errors that may remain in the article are my fault, and not the fault of any of my reviewers." (Meredith Angwin)


I would like to express my appreciation for Meredith's efforts in writing this piece for two reasons. First, this piece is timely; the problems with the replacement S/G's at San Onofre are very "now" in the news, and readers really do want to know what's going on. There is some considerable weight given in Mr. Gundersen's work to the notion that not enough work is / was done to design replacement steam generators; Meredith's piece puts that to bed in short order. The second reason is that this piece goes right toward APR's stated focus of presenting the inside details on the technology of nuclear energy and the history of that technology. It is rare that we get this sort of inside glimpse at some of the work done in the past during this whole long course of the development of nuclear energy, and we're all the better for it having seen Meredith's work reproduced above.

If any of the terms or mechanical details above were confusing please remember to see this post on APR which gives a great deal of detail on the overall design of steam generators.

Thanks again to Meredith and to all of the people who reviewed and commented on this work prior to its publication here on APR.

9:20 PM Eastern 4/24/2012


  1. Meredith,
    A great piece of writing that shows scare-mongering by self-proclaimed, agenda-based, experts doing reports for a fee can be refuted by good reasoning.

    The US nuclear industry has THE HIGHEST CAPACITY FACTOR IN THE WORLD. That fact alone indicates that nearly all reactors (their original designs had shortcomings) have been upgraded with the latest replacements parts to ensure such high CFs.

  2. Thank you Willem! Yes, the U.S. nuclear industry has the best capacity factors in the world. That is something to be proud of! Glad you enjoyed the article. Will tells me it is getting a lot of hits.

  3. Very informative article!

    I'm just wondering, granted I'm no engineer, but instead of tubes why not a very large hundreds-layer sandwich of "heat exchange" plates creating flat fluid "channels", each other layer carrying secondary and primary water via eaches own inlet outlets? Wouldn't such an increase of area over tubes be superior, and wouldn't it get even more efficient the thinner (as you dare) the layers become? It might not even require the same pressure needed for the tube system, but just thinking over my head after looking at water trapped 'tween a couple layers of aluminum foil here!

    James Greenidge
    Queens NY

  4. @jim: Two words - flow reversal!

  5. Very good article Meredith.
    Carefully treading where the angels fear, The root cause may have been the specification that the thermal capacity of the RSG be the same as the OSG with a lifetime to 2020.

    I understand that 690 is superior in every way except heat transfer. The like with like trade off may have been to run the identical mechanical design with improved materials and fabrication at LOWER power (no additional tubes, same number of plates). Put another way, the originally committed capacity AND lifetime may not have been achievable with this reactor system design. The improvements in materials and fabrication of MHI's RSG may have lead to the lifetime being meet, but with the constraint of identical mechanical design, not at the specified capacity.

    An alternative design trade off might have been to use improved fabrication with 600 to achieve full capacity with a somewhat improved lifetime within the identical mechanical configuration.

    A take away from the SONGS experience for the engineering community might be to be skeptical that all the corners of the given specification are achievable.

    A take away for the operating community may be that when told there is risk in all the specifications being meet, compromising on the economic specs (power/capacity) may be the most financially prudent.